This case study will explore how we helped a prominent AWS reseller with an active cost optimization strategy achieve a 164% savings increase in their first month of service.
Before we begin working with a new customer, we perform a free AWS Compute Savings Analysis to summarize their historical savings performance, overlay our algorithm on their data, and quantify the opportunity for optimization. That analysis is self-serve via our website and takes 5 minutes to configure and 45 minutes to review the results.
A history of lost savings and missed opportunities
The customer in this case study is an AWS reseller with multiple payer accounts. The reseller knew there was an opportunity to more aggressively use savings instruments to increase their margin, but were unsure how to do so without increasing their business risk or losing money.
The business risk resellers take when trying to expand their margins with savings instruments is particularly complicated because the reseller never controls or knows the usage patterns of the underlying end customers. If the reseller commits too aggressively and the end customers drop their usage, the reseller is stuck with losses from unutilized commitments. If the reseller commits too conservatively, they leave margin on the table. They reached out to us for help in July of 2020 to help them better manage that risk and optimize savings.
In June of 2020, the total spend for the payer account below was just above $400,000, broken down in Figure 1 by On-demand Instances (Figure 1, red), Reserved Instances purchased by end-customers (Figure 1, light purple), Reserved Instances purchased by the reseller (Figure 1, dark purple), Savings Plans purchased by end-customers (Figure 1, magenta), along with a smattering of Fargate, Lambda, and EC2 Spot.
After excluding the commitments made by end customers for their own cost savings benefit, there was a remaining spend base of $280,000 per month to generate margin against. Previously, the reseller’s margin optimization strategy involved a small pool of about $45,000 in Standard Reserved Instances (RIs). This was producing mixed results, with chronic RI underutilization costing the customer significant financial loss.
In June of 2020, for example, the reseller had saved $8,000 (Figure 2, green bar), but missed out on an incremental $17,000 due to unutilized RIs (Figure 2, red bar). Looking back over the preceding 12 months, the reseller generated, on average, less than $10,000 per month in savings when they should have been capturing more than $20,000 per month in savings.
Uncovering the Effective Savings Rate
ProsperOps’ core output metric is called the Effective Savings Rate (ESR) which calculates the overall discount rate received on the global compute spend.
Every business leveraging AWS savings instruments can benefit from using the Effective Savings Rate to determine their ROI. ESR takes into account all elements of financial optimization: the utilization of savings instruments, the coverage of commitments relative to usage, and which discount rates are being applied to the commitment (which vary by instance, region, operating system, etc.).
To learn more about how we calculate ESR, and why it’s more reliable than other AWS financial metrics, read our three-part blog series here.
We found the reseller had an average ESR of just 2%, placing them in the 53rd percentile of AWS optimizers (Figure 2, yellow line). For reference, the 90th percentile of AWS optimizers achieve an ESR of 32%.
ProsperOps delivers immediate and dramatic savings
To begin, our service deploys a conservative and full-term “base” commitment of Compute Savings Plans to cover their EC2, Fargate, and Lambda needs, as well as smaller amounts of EC2 usage in non-core regions (Figure 3, green line).
The service then augments the immutable commitments of the base layer with a dynamic, “flex” layer of short-term Convertible RIs (Figure 3, light green line) to collectively cover 95% of overall usage at near 100% utilization. Because our algorithms and automation adjust the flex commitment to track usage up or down in real-time, the reseller can maximize utilization, coverage, and discount, without having any control over or knowledge of the underlying usage base.
Compare the ProsperOps approach with the reseller’s historical commitment strategy. The chart below shows their commitment over the prior twelve months, with the most recent commitment (Figure 3, red line) representing only 22.4% of usage (Figure 3, purple bars). When you only account for the utilized portion of the commitment that generates a discount (Figure 3, gray line) the coverage drops to 16.2% of usage, meaning 83.8% of the usage was paid at the on-demand rate.
Within just one month of implementing our service, the reseller saw an immediate 164% increase in their net savings (Figure 4, green bar), accompanied by a dramatic 8% increase in their ESR (Figure 4, yellow line).
Those outcomes are net of the ProsperOps charge and we expect to drive the reseller’s ESR towards the 98th percentile, which equates to discounts of 40+%.
Best of all, this work was handled automatically by ProsperOps, with no manual intervention by the reseller’s team. This allows them to focus on other issues with confidence that their AWS spend is optimized for both financial performance and stability.
More savings. Less commitment risk. No manual work. No rearchitecture. A larger cloud budget. One less thing to worry about. If you’d like to understand how ProsperOps could generate similar results for your business, signup to receive a complimentary AWS Compute Savings Analysis.